基于间隙度的无人机林地航拍图像序列拼接方法

发布时间：2018-04-14 09:05

本文选题：图像拼接 + 无人机图像序列　；参考：《北京林业大学学报》2017年06期

【摘要】：无人机林地航拍图像具有的分辨率高、数据量大、边缘丰富的特点,造成了特征点配准中误匹配率的增加,因此本文提出了一种新的无人机林地航拍图像序列拼接方法。分形中的间隙度可用于描述图像区域纹理的粗糙程度,本文首先利用间隙度特征选取图像中局部图像块作为候选区域查找特征点,减少了待配准的特征点数量,提高了特征点配准正确率;其次,采用全局拼接技术变换图像,减少传统拼接中矩阵连乘产生的误差的积累和传播。实验中选取了不同拍摄高度的两组图像序列,将本方法与传统的全局SURF特征方法和降采样图像拼接方法进行了对比,结果显示本方法可以有效拼接图像,同时不会损失原图像的精度信息,并从视觉效果和均方根误差两个角度证明了本文方法优于其他两种方法。
[Abstract]:Aerial photography images of UAV forest land have the characteristics of high resolution, large amount of data and abundant edges, which result in the increase of mismatch rate in feature point registration. Therefore, a new method of aerial photography image sequence mosaic for UAV forest land is proposed in this paper.The gap degree in the fractal can be used to describe the roughness of the texture of the image region. Firstly, the gap degree feature is used to select the local image blocks in the image as the candidate region to find feature points, which reduces the number of feature points to be registered.The accuracy of feature points registration is improved. Secondly, the global stitching technique is used to transform images to reduce the accumulation and propagation of errors caused by matrix concatenation in traditional stitching.In the experiment, two groups of image sequences with different shooting heights are selected and compared with the traditional global SURF feature method and the de-sampling image mosaic method. The results show that the proposed method can effectively concatenate images.At the same time, the accuracy information of the original image will not be lost, and it is proved that the proposed method is superior to the other two methods in terms of visual effect and root-mean-square error.
【作者单位】：北京林业大学信息学院;
【基金】：教育部高等学校优秀青年教师研究基金项目(2015ZCQ-XX)
【分类号】：TP751

【相似文献】